Risk factors associated with greater trochanteric overgrowth secondary to Legg-Calvé-Perthes disease: a retrospective observational study

Abstract

Background

Legg-Calve’-Perthes disease (LCPD) is a common osteoarthrosis in children. Studies on LCPD are mainly focused on how to maintain the sphericity of the femoral head and improve the congruency of the hip, yet there has not been enough attention given to secondary deformities such as greater trochanteric overgrowth (GTO). Therefore, we conducted a retrospective study to determine the risk factors associated with GTO secondary to LCPD.

Methods

The demographic and imaging data of children with LCPD from 2008 to 2021 were retrospectively analyzed. The age at diagnosis, sex, lateral type, treatment methods, Herring lateral pillar classification, Catterall classification, the signs of hip“at risk” and Stulberg classification were recorded. The articulotrochanteric distance (ATD) value of hip < + 5 mm was diagnosed as GTO. Spearman rank correlation test and logistic regression analysis were used to evaluate possible correlations and the risk factors for GTO secondary to LCPD.

Results

125 children with LCPD were enrolled, 91 patients without GTO and 34 patients with GTO, the positive rate of GTO was 27.2%. Spearman rank correlation test showed that GTO was correlated with Herring lateral pillar classification(ρ = 0.276, p = 0.002), the sign of hip “at risk” (ρ = 0.305,p<0.001) and Stulberg classification ( ρ = 0.346, p<0.001). Multivariate logistic regression analysis suggested that the risk of GTO was 2.7 times for LCPD with the sign of hip “at risk” than without the sign of hip “at risk” (OR:2.672,95%CI:1.030–6.930), 2.9 times (OR:2.941, 95%CI:1.181–7.324) after combined sub-classifications of similar properties. The risk of GTO was 3.9 times for LCPD with Stulberg III/IV/V than with Stulberg I/II (OR:3.899, 95%CI:1.481–10.262).

Conclusion

Hip ‘at-risk’ signs are independent risk factors with early predictive value for GTO secondary to LCPD. The Stulberg classification is positively correlated with GTO at skeletal maturity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-06148-9.

Background

LCPD refers to idiopathic epiphyseal necrosis of the femoral head in children, which can lead to deformity of the femoral head and early degenerative hip arthritis with a high disability rate [1, 2]. The pathophysiological processes of LCPD are the disturbance of blood supply, which is caused by various factors, leading to avascular epiphyseal necrosis of the femoral head, followed by the recovery of the blood supply and reconstruction of the epiphysis of the femoral head [3]. The clinical prevention and treatment of LCPD mainly focus on how to restore the sphericity of the femoral head and reduce the occurrence of hip degenerative arthritis but do not pay enough attention to secondary deformities, such as GTO [4, 5].

GTO is a common but easily overlooked residual deformity of LCPD, with an unclear pathological mechanism. It often results in hip abductor insufficiency, pelvic tilt, fatigue, walking pain, and a positive Trendelenburg sign [6].The ATD has previously been the most valuable index to judge GTO. The diagnostic threshold of ATD < + 5 mm was adopted based on established literature [7, 13, 14], as this value is associated with clinically significant abductor insufficiency and gait impairment.When ATD is less than + 5 mm, patients can be diagnosed with GTO [6, 7].Although GTO may be treated with surgical methods such as distal transfer of the greater trochanter and greater trochanteric apophyseodesis, the therapeutic effect is controversial [6, 8], therefore it is necessary to determine the risk factors associated with GTO and early preventive treatment to reduce its occurrence. At present, there are few reports on the risk factors for GTO [7, 9]. Through long-term clinical observation, we found that GTO might be associated with the degree of epiphyseal necrosis of the femoral head.

In this study, we retrospectively reviewed the demographic and imaging data of LCPD patients who were diagnosed, treated and followed up in our institution and analyzed the related risk factors for GTO secondary to LCPD.

Methods

Demographic data

The demographic and imaging data of patients who were diagnosed as LCPD, treated and followed-up were retrospectively analyzed in our institution from 2008 to 2021.The study was conducted according to the ethical principles stated in the Declaration of Helsinki. The study received approval from the Institutional Review Board/Ethics Committee of affiliated hospital of Zunyi medical university (Ethical review approval number: KLL-2022-450). Inclusion criteria: (1) diagnosed clearly as LCPD according to clinical manifestations and imaging findings; (2) followed-up to skeletal maturity;3. had available demographic and imaging data. Exclusion criteria: 1.demographic and imaging data were not incompleted; 2. epiphyseal necrosis of the femoral head caused by developmental dysplasia of the hip, femoral neck fracture, achondroplasia, or other clear causes. The age at diagnosis as LCPD, sex, lateral type, treatment methods were recorded. The treatment methods were divided into conservative and operative treatments according to the age and the severity of epiphyseal necrosis of the femoral head [2].Conservative treatment included regular follow-up, adductor dissection, non-weight bearing brace in abduction of hip, and plaster fixation in abduction and neutral rotation of hip. Operative treatment included pelvic osteotomy and proximal femoral varus osteotomy, which was designed to increase coverage of the femoral head by the acetabulum [10].

Imaging data

All cases underwent routine pelvic anteroposterior and Lauenstein radiographs before treatment, and MRI of the hip was performed when the diagnosis was suspected. Follow-up was performed every 3 months for two years after the initial diagnosis. The ATD value is the distance between the two lines parallel to Hilgenreiner’s line passing through the tip of the greater trochanter and the proximal point of the femoral head. Consistent with prior studies [7, 13, 14], GTO was diagnosed when ATD < + 5 mm – a threshold validated for its association with functional impairment (abductor weakness and abnormal gait).This threshold (ATD < + 5 mm) was selected based on its established clinical relevance: Kitoh et al. demonstrated that this cutoff correlates with clinically significant abductor insufficiency and gait impairment in LCPD patients [7], while Leitch et al. confirmed its association with functional deficits such as Trendelenburg’s sign [14]. Akpinar et al. further validated its use in identifying trochanteric overgrowth requiring intervention [13], supporting its reliability as a diagnostic criterion.A positive ATD value indicated that the greater trochanter was lower than the femoral head, and a negative value indicated that the femoral head was lower than the greater trochanter (Fig. 1). The severity of femoral head epiphyseal necrosis was assessed using the Herring lateral pillar classification, Catterall classification, and hip ‘at-risk’ signs—including femoral head eccentricity, lateral subluxation, metaphyseal cysts, or lateral calcifications.Each case was evaluated according to the most serious type in the course of the disease.Stulberg classification was used to evaluate the spherical structure of the femoral head and its congruency with the acetabulum at skeletal maturity [11, 12].

Fig. 1.

Pelvic anteroposterior radiographs of LCPD. A represents the affected hip, which was diagnosed as GTO, the ATD value was −8 mm. B represents the normal hip, the ATD value was 21 mm

Statistical analysis

All statistical analyses were performed using SPSS 27.0 (SPSS, Chicago, USA). Sample size calculation was conducted prior to data analysis using G*Power 3.1, based on an anticipated GTO incidence of 22–25% (Kitoh et al., 2013), α = 0.05 (two-tailed), power = 80%, and an expected effect size (OR = 2.5). This yielded a minimum required sample size of 116 patients. Our final cohort of 125 patients met this requirement.The continuous variables were expressed as the mean ± standard deviation, the normally distributed data were compared by test, the non-normally distributed data were compared by the nonparametric Mann-Whitney U test, and the counting variables were compared by the chi-square test or Fisher’s exact test. Spearman rank correlation test and Logistic regression analysis were used to evaluate possible correlations and the risk factors for GTO. Stulberg classification, which reflects femoral head morphology at skeletal maturity, is inherently an outcome measure rather than a pre-existing risk factor. Consequently, it was not included in regression models as a predictor. Instead, its association with GTO was evaluated solely through correlation analysis to explore shared morphological outcomes.The stepwise regression method was used to find the appropriate logistical model, multivariable logistic regression analysis to estimate OR and 95% CI of having GTO when the covariates were significant at p-values < 0.05 in the univariate logistic regression.To evaluate the reliability of the logistic regression models, the following diagnostics were performed: Discriminative ability: The Area Under the Curve (AUC) was 0.72 (95% CI: 0.63–0.81), indicating moderate capacity to distinguish GTO cases from non-cases.Calibration: The Hosmer-Lemeshow test showed good agreement between predicted and observed probabilities (χ²=6.23, df = 8, p = 0.62), confirming adequate model calibration.”

Results

Tables 1 shows that variables such as sex (male: 86.8% in non-GTO vs. 82.4% in GTO, p = 0.729), lateral type (left: 53.8% vs. 55.9%, p = 0.839), and treatment methods (conservative: 58.2% vs. 47.1%, p = 0.263) did not differ significantly between the two groups, indicating they were not associated with GTO. These non-contributory variables are included for completeness but were not further analyzed due to their lack of statistical significance.

Table 1.

The demographic and imaging variables of 125 cases with LCPD

Characteristics	Patients without GTO (n = 91)	Patients with GTO (n = 34)	p value
ATD value,mm	14.53 ± 4.41	−3.18 ± 6.09
Age at diagnosis, years	7.97 ± 1.97	8.17 ± 1.70	0.552*
Sex, male	79(86.8)	28(82.4)	0.729
Lateral type, left	49(53.8)	19(55.9)	0.839
Herring classfication			0.005
A	12(13.2)	3(8.8)
B	41(45.1)	8(23.5)
BC	27(29.7)	9(26.5)
C	11(12.1)	14(41.2)
Catterall classfication			0.800
I	9(9.9)	3(8.8)
II	37(40.7)	11(32.4)
III	32(35.2)	15(44.1)
IV	13(14.3)	5(14.7)
Stulberg classfication			<0.001
I + II	78(85.7)	18(52.9)
III	8(8.8)	9(26.5)
IV + V	5(5.5)	7(20.6)
The sign of hip “at risk”,yes	26(28.6)	21(61.8)	<0.001
Treatment methods, conservation	53(58.2)	16(47.1)	0.263

*Mann-Whitney U test.

According to conventional interpretive criteria for correlation coefficients, these associations are considered weak-to-moderate: Herring classification (ρ = 0.276) and the sign of hip “at risk” (ρ = 0.305) show mild-to-moderate correlations, while Stulberg classification (ρ = 0.346) indicates a slightly stronger but still moderate relationship with GTO[18].In univariate logistic regression analysis, age at diagnosis, sex, lateral type, Catterall classification and treatment methods did not significantly correlate to GTO.The Herring lateral pillar classification and hip ‘at-risk’ signs were identified as potential risk factors for GTO. Stulberg classification was excluded from regression modeling as it constitutes an outcome measure rather than a predictor variable.The risk of GTO was five times in Herring C than in Herring A (OR:5.091, 95%CI:1.146–22.620), four times more likely to occur with the sign of hip “at risk” than without the sign of hip “at risk” (OR:4.038,95%CI:1.765–9.241), patients with Stulberg III (OR:4.875, 95%CI:1.653–14.377) and Stulberg IV/V(OR 6.067, 95%CI:1.726–21.321) were respectively five times and six times more likely to developing GTO than Stulberg I/II at skeletal maturity. Multivariate logistic regression analysis with unmerged sub-classifications (Supplementary Table S1) identified the sign of hip “at risk” as an independent risk factor for GTO (p = 0.043), with a 2.7-fold increased risk (OR:2.672, 95%CI:1.030–6.930). However, due to small sample sizes in individual sub-groups (e.g., Herring A: 12 non-GTO vs. 3 GTO cases), some estimates showed wide confidence intervals, indicating limited statistical stability (Table 2).

Table 2.

Spearman rank correlation test for GTO in 125 LCPD patients 

Characteristics	ρ	p
Age at diagnosis	0.053	0.554
Sex	0.057	0.531
Lateral type	−0.018	0.840
Herring classfication	0.276	0.002
Catterall classfication	0.064	0.476
The sign of hip “at risk”	0.305	<0.001
Stulberg classfication	0.346	<0.001
Treatment methods	0.100	0.267

To address this, sub-classifications with similar properties were merged (Herring A + B, BC + C; Catterall I + II, III + IV; and Stulberg III + IV/V) based on shared pathological and prognostic similarities. Given the small sample sizes in specific sub-classifications, individual subgroups lacked sufficient statistical power to yield stable estimates, increasing the risk of imprecise associations. Merging was performed based on shared pathological and prognostic similarities: For Herring classification: A and B represent mild-to-moderate involvement with preserved lateral pillars, while BC and C indicate severe collapse [7, 12].For Catterall classification: I/II (limited necrosis) and III/IV (extensive necrosis) align with the original intent to distinguish necrosis extent [11].For Stulberg classification: III/IV/V collectively indicate abnormal sphericity, distinct from I/II [12].A sensitivity analysis confirmed key associations (e.g., “at risk” sign with GTO) remained consistent in merged and unmerged models, ensuring no meaningful distinctions were obscured.

This merged analysis (Table 3) served as the primary model due to improved stability, confirming the sign of hip “at risk” (OR:2.941, 95%CI:1.181–7.324) and Stulberg III/IV/V (OR:3.899, 95%CI:1.481–10.262) as significant factors. Unmerged results (Supplementary Table S1) are provided for transparency, showing consistent trends but lower stability due to smaller sub-group sizes.

Table 3.

Logistic regression analysis after combined sub-classification for GTO in 125 LCPD patients

Characteristics	B	SE	Wald χ2	p	OR	95% CI	B	SE	Wald χ2	p	OR	95% CI
Herring classfication(BC + C/A + B)	1.070	0.424	6.379	0.012	2.916	1.271–6.692	0.335	0.492	0.463	0.496	1.397	0.533–3.662
Catterall classfication(III + IV/I + II)	0.379	0.407	0.867	0.352	1.460	0.658–3.241
The sign of hip “at risk”(yes/no)	1.396	0.422	10.922	<0.001	4.038	1.765–9.241	1.079	0.465	5.370	0.020	2.941	1.181–7.324
Stulberg classfication(III + IV + V/I + II)	1.674	0.456	13.485	<0.001	5.333	2.183–13.032	1.361	0.494	7.593	0.006	3.899	1.481–10.262

Model diagnostics for the merged sub-classification model: AUC = 0.72 (95%CI:0.63–0.81), Hosmer-Lemeshow test p = 0.62, pseudo R²=0.18

Discussion

The pathogenesis and influencing factors of GTO secondary to LCPD are unclear, GTO can lead to hip abductor insufficiency, pelvic instability, and limping gait. The radiographic measurements for evaluating GTO include the ATD value, the ratio of the ATD value of the affected hip and normal hip, the greater trochanter height and the physeal-trochanteric distance, but the ATD value is the most valuable. The cases can be diagnosed as GTO when ATD < 5 mm [13]. Kitoh et al. found that the ratio of ATD values of the affected hip to the normal hip in LCPD was 57%, and the ATD was less than + 5 mm in 22% of LCPD patients [7]. Leitch et al. reported that the incidence of GTO with an ATD < + 5 mm was 23% [14]. Similarly, 34 cases (27.2%) were diagnosed as GTO in 125 LCPD children in this retrospective study, and the ratio of the ATD value of the affected hip to that of the normal side was 45.4%. In summary, LCPD can reduce the ATD value of the affected hip to varying degrees, and GTO may be a common and inevitable complication of LCPD, but it is not clear which risk factors affect the occurrence of GTO.

As the epiphyseal development of the femoral head and the greater trochanter is coordinated, we speculated that the severity of the epiphyseal necrosis of the femoral head may be the related factors of GTO secondary to LCPD [8, 15].Kim et al. made the LCPD animal model of piglets and found that cystic lesions of the proximal femur led to the growth impairment of the proximal femur, which affected the developmental of the femoral head [16].Kitoh et al. clinical studies confirmed that the incidence of GTO in Herring C was significantly higher than that in Herring B/BC, the more serious the epiphyseal collapse of the femoral head, the higher the incidence of GTO secondary to LCPD [7]. Wang et al. reported a higher incidence of GTO in patients with hip ‘at-risk’ signs, such as complete epiphyseal collapse and metaphyseal cystic lesions [9].This study suggests that the sign of hip “at risk” is an independent risk factor with early predictive value for GTO secondary to LCPD. Stulberg classification shows significant correlation with GTO at skeletal maturity, reflecting their shared status as late-stage morphological outcomes.Finally, it was confirmed that the sign of hip “at risk” was the independent risk factor for GTO by multivariate logistic regression analysis, the risk of developing GTO in children with LCPD with the sign of hip “at risk” is 2.7 times than without the femoral head risk sign(OR:2.672, 95%CI:1.030–6.930).Considering the small sample size of sub-classifications in the covariates, after combining the partial sub-classifications, the sign of hip “at risk” was still an independent risk factor for GTO secondary to LCPD in univariate and multivariate analysis.Stulberg classification, assessed at skeletal maturity, reflects the final morphological outcome of femoral head development rather than a pre-existing risk factor for GTO. Its significant correlation with GTO (ρ = 0.346, p<0.001) likely arises from shared pathological processes (e.g., impaired epiphyseal remodeling) that contribute to both femoral head deformity and trochanteric overgrowth, rather than a causal relationship. This reinforces the need to interpret Stulberg classification as a correlate of GTO, not a predictive factor. GTO secondary to LCPD can be treated with distal transfer of the greater trochanter [6], which aims to increase gluteus medius tension, improve the altered hip biomechanics, and reduce limping gait. For children with sufficient growth potential, greater trochanteric epiphysis fusion or epiphyseodesis can be used to prevent the occurrence of GTO [8, 13]. Shah H et al.reported that the positive Trendelenburg’s sign was less common in children who underwent trochanteric epiphysis fixation than in children who did not undergo surgery, but overcorrection was also found in 10% of children after routine prophylactic trochanteric epiphyseal fixation [17].So early intervention with greater trochanteric epiphysis fusion or epiphyseodesis in children with LCPD who have risk factors for GTO can accurately prevent the occurrence of GTO after skeletal maturation.There are some limitations in this study. Firstly, the overall small sample size, especially patients with GTO, may affect the accuracy of the results. Second, imaging signs (e.g., hip ‘at-risk’ signs) were assessed using Catterall’s established criteria, which include femoral head eccentricity, lateral subluxation, metaphyseal cysts, and lateral calcifications [11]. However, interobserver reliability was not formally evaluated (e.g., using kappa coefficients) to quantify consistency between raters, which may introduce subjective bias. Thirdly, cases with missing or incomplete data (e.g., insufficient follow-up imaging or incomplete demographic records) were excluded from analysis (n = 18), potentially introducing selection bias as these cases may differ in disease severity or outcomes. Fourthly, imaging evaluations (e.g., ATD measurements and classification assessments) relied on manual measurements, which may be subject to intra- and interobserver variability despite adherence to standard protocols. Fifthly, although this study did not indicate a correlation between treatment methods and GTO, proximal femur varus osteotomy has been associated with an increased risk of GTO in previous reports [6]. Sixthly, the study includes data spanning 2008–2021. While older data were essential to ensure complete follow-up to skeletal maturity (critical for assessing Stulberg classification and GTO), minor variations in imaging techniques and clinical management protocols over this period may have introduced subtle inconsistencies, which should be considered when interpreting results.Seventhly, cases with missing or incomplete data (e.g., insufficient follow-up imaging or incomplete demographic records) were excluded from analysis (n = 18), potentially introducing selection bias as these cases may differ in disease severity or outcomes.Finally, this retrospective design precluded control of unmeasured confounding factors (e.g., nutritional status or physical activity levels), and prospective multicenter studies with standardized imaging assessment protocols are needed to validate our findings.

Conclusions

This study suggests that the sign of hip “at risk” is an independent risk factor with early predictive value for GTO secondary to LCPD. Stulberg classification shows significant correlation with GTO at skeletal maturity, reflecting their shared status as late-stage morphological outcomes.

Abbreviations

LCPD

Legg-Calve'-Perthes disease

GTO

Greater trochanteric overgrowth

ATD

Articulotrochanteric distance

OR

Odds ratios

CI

Confidence intervals

Authors’ contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by KZC, TJZ. The first draft of the manuscript was written by KZC, TJZ, all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by Guizhou Province’s Science and Technology Plan Project [Basis (2018) 1430 and Support (2021) General 076].

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of affiliated hospital of Zunyi medical university (Ethical review approval number: KLL-2022-450). And Informed consent was obtained from the parents of all individual participants included in the study.

Consent for publication

The authors affirm that human research participants provided informed consent for publication of the images in Figure.

Competing interests

The authors declare no competing interests.